daBongo LMS AI Training Courses

Lesson Content

ChatGPT as a learning partner.

One of ChatGPT's most valuable capabilities is as a personalized explainer. Unlike a search engine that returns existing articles, ChatGPT can:

• Explain concepts at exactly your level of background knowledge
• Answer the specific follow-up question that occurred to you
• Use analogies from your field or experience
• Test your understanding with tailored questions
• Adjust the explanation when it misses the mark

To use ChatGPT as a learning partner effectively:

"I want to understand [concept]. I have a background in [your background] but no experience with [new topic]. Explain it in a way that builds from what I already know. After you explain, give me three questions to test my understanding â€" ask them one at a time."

Why ChatGPT can be confidently wrong.

Understanding this is the most important safety concept for new users.

ChatGPT generates text that is statistically likely to follow from what came before â€" it is trained to produce plausible, coherent, well-organized language. But "plausible and well-organized" is not the same as "accurate." ChatGPT can produce:

• Hallucinated facts: Statistics, citations, quotes, and details that sound real but are fabricated
• Outdated information: Its training data has a knowledge cutoff â€" events after that date are not in its knowledge base
• Confident errors: Mistakes expressed with the same tone and structure as accurate information

This is the most important thing to internalize: ChatGPT's confidence level does not indicate accuracy. A confidently stated hallucination reads exactly like a confidently stated fact.

The four-tier verification framework.

Match your verification effort to the stakes of the decision:

Tier 1 â€" Low stakes (plausibility check): Mistakes are easily corrected with little consequence. *Examples: brainstorming, casual learning, organizing your thoughts, drafting internal notes.* Verify: Does this seem plausible? Does it align with what you know? A quick sanity check is sufficient.

Tier 2 â€" Medium stakes (spot-check important claims): Mistakes could cause meaningful cost, wasted time, or embarrassment. *Examples: client-facing documents, information you will share publicly, research summaries you will act on.* Verify: Check the most important facts and claims against authoritative sources. Do not accept statistics, citations, or specific claims at face value.

Tier 3 â€" High stakes (verify all claims): Mistakes could cause significant professional, financial, or reputational harm. *Examples: financial analysis, legal documents, medical information, information being presented to leadership.* Verify: Every material claim verified with an authoritative source. Do not act on ChatGPT's output without formal verification.

Tier 4 â€" Non-negotiable (professional review required): Mistakes could cause irreversible harm â€" legal, medical, financial, safety-critical. *Examples: medical treatment decisions, legal filings, financial investment advice, safety engineering.* Verify: Licensed professional review is required regardless of how authoritative ChatGPT's response appears.

Applying the framework.

Before using any ChatGPT output, ask: "What is the worst realistic outcome if this is wrong?" Your answer tells you which tier applies.